grade expectations: the effects of expectations on fairness and satisfaction perceptions

21
Journal of Behavioral Decision Making J. Behav. Dec. Making, 16: 375–395 (2003) Published online 24 September 2003 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/bdm.452 Grade Expectations: The Effects of Expectations on Fairness and Satisfaction Perceptions BENNETT CHERRY 1 *, LISA D. ORDO ´ N ˜ EZ 2 and STEPHEN W. GILLILAND 2 1 California State University, San Marcos, CA, USA 2 The University of Arizona, USA ABSTRACT Previous research has documented the impact of self-derived expectations as reference points in the evaluation of outcomes (e.g. Ordo ´n ˜ez, Connolly, & Coughlan, 2000; van den Bos et al., 1998). In the present paper we extend these studies by investigating the effects of individuals’ performance expectations on their subsequent evaluations of per- sonally-relevant outcomes. In three separate studies, both in the laboratory and in the field, students’ actual grade outcomes fell short, met, or exceeded grade expectations. From this information, the students evaluated their fairness and satisfaction with the actual grade outcome. The studies provide complementary results that distinguish fair- ness and satisfaction as different constructs based on the impact of expectations on eva- luations of actual outcomes. Results demonstrate that expectations are important to perceptions of fairness and are less important to perceptions of satisfaction. Fairness judgments appear to be governed by an expectation matching proposition; whereby if the expectation is met, the outcome is fair. Whereas, satisfaction judgments are deter- mined by the value of the actual outcome to the individual. Participants also evaluate the fairness of outcomes differently using hypothetical scenarios than they do when they experience actual outcomes in natural contexts. Copyright # 2003 John Wiley & Sons, Ltd. key words fairness; satisfaction; expectation; grades; perceptions In many of the daily events of our lives (work, school, sports activities), we receive evaluations from others of our performance relative to some criterion. In the classroom, for example, teachers assign exam grades as a measure of student learning. In most cases, the student usually has some expectation as to the likely out- come of the evaluation. An interesting question is, ‘What happens when the outcome either fails to meet or exceeds our expectation?’ Social scientists have examined the comparisons that a person makes with others in an effort to determine what motivates people and what components of satisfaction are guided by these social comparisons (Adams, Copyright # 2003 John Wiley & Sons, Ltd. *Correspondence to: Bennett Cherry, California State University, San Marcos, College of Business Administration, San Marcos, CA 92096–0001, USA. E-mail: [email protected]

Upload: bennett-cherry

Post on 06-Jun-2016

217 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Journal of Behavioral Decision Making

J. Behav. Dec. Making, 16: 375–395 (2003)

Published online 24 September 2003 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/bdm.452

Grade Expectations: The Effects ofExpectations on Fairness andSatisfaction Perceptions

BENNETT CHERRY1*, LISA D. ORDONEZ2 and STEPHEN W. GILLILAND2

1California StateUniversity, SanMarcos, CA, USA2TheUniversityofArizona, USA

ABSTRACT

Previous research has documented the impact of self-derived expectations as referencepoints in the evaluation of outcomes (e.g. Ordonez, Connolly, & Coughlan, 2000; vanden Bos et al., 1998). In the present paper we extend these studies by investigating theeffects of individuals’ performance expectations on their subsequent evaluations of per-sonally-relevant outcomes. In three separate studies, both in the laboratory and in thefield, students’ actual grade outcomes fell short, met, or exceeded grade expectations.From this information, the students evaluated their fairness and satisfaction with theactual grade outcome. The studies provide complementary results that distinguish fair-ness and satisfaction as different constructs based on the impact of expectations on eva-luations of actual outcomes. Results demonstrate that expectations are important toperceptions of fairness and are less important to perceptions of satisfaction. Fairnessjudgments appear to be governed by an expectation matching proposition; wherebyif the expectation is met, the outcome is fair. Whereas, satisfaction judgments are deter-mined by the value of the actual outcome to the individual. Participants also evaluatethe fairness of outcomes differently using hypothetical scenarios than they do whenthey experience actual outcomes in natural contexts. Copyright # 2003 John Wiley& Sons, Ltd.

key words fairness; satisfaction; expectation; grades; perceptions

In many of the daily events of our lives (work, school, sports activities), we receive evaluations from others

of our performance relative to some criterion. In the classroom, for example, teachers assign exam grades as

a measure of student learning. In most cases, the student usually has some expectation as to the likely out-

come of the evaluation. An interesting question is, ‘What happens when the outcome either fails to meet or

exceeds our expectation?’

Social scientists have examined the comparisons that a person makes with others in an effort to determine

what motivates people and what components of satisfaction are guided by these social comparisons (Adams,

Copyright # 2003 John Wiley & Sons, Ltd.

* Correspondence to: Bennett Cherry, California State University, San Marcos, College of Business Administration, San Marcos, CA92096–0001, USA. E-mail: [email protected]

Page 2: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

1965). All comparison theories (e.g. equity theory; social comparison theory) rely on a referent other, some-

one with whom to compare oneself. While much has been written about these social-comparison based refer-

ence points, less attention has been given to nonsocial comparisons as reference points. In Kulik and

Ambrose’s (1992) discussion of the determinants of choosing a referent other, they offer a possibility of self

as a nonsocial referent. Moreover, Gilliland (1993) mentions that self as a referent can create a performance

expectation. This internal reference point is suggested to be used to evaluate outcomes.

Past research demonstrates that these internal reference points influence evaluations of outcomes

(e.g. Ordonez, Connolly, & Coughlan, 2000; van den Bos et al., 1998). Austin and Walster (1974) provided

evidence that expectancies can influence an individual’s satisfaction and fairness judgments of outcomes—if

an individual expects inequity to occur, the distress is not as great as when inequity is not expected. More

recently, Gilliland (1994) found that participants’ expectations of their chances of being hired were

positively related to their perception of fairness in a hiring situation. The importance of expectations has

also been found in the marketing domain with respect to post-purchase product evaluation (e.g. The Expec-

tancy Disconfirmation Model of Satisfaction from Oliver, 1997). From the job-satisfaction literature, the

concept of met expectations has been reviewed and meta-analyzed by Wanous et al. (1992). In their

review, Wanous and colleagues found that individuals’ job-related expectations, whether met or unmet, influ-

enced organizationally-relevant outcomes such as job satisfaction, organizational commitment, intentions to

leave, job survival, and job performance. However, recent research suggests that met expectations are not

as important in determining job satisfaction as are an individual’s post-entry experiences (e.g. Hom

et al., 1999).

In the present paper, we investigate the effect that these self-generated expectations have on two

important subjective evaluations of outcomes: fairness and satisfaction. We do not focus on how expecta-

tions are formulated or the confidence level expressed in these expectations. Instead, we investigate the

extent to which an individual’s expectations of a future outcome influence her subsequent feelings related

to that outcome. In the present paper, we have chosen grades assigned in a college class setting for investi-

gating these judgments. Exam grades are important outcomes for our student participants and, as such, the

outcome evaluations should be authentic due to their familiarity with the context and the items under inves-

tigation. Our paper has two main goals: (1) to investigate how the expectations of grades affect fairness

and satisfaction ratings, and how these two measures relate to one another; and (2) to examine the relation-

ship between responses to hypothetical scenarios and the more authentic actual experienced outcomes in

the classroom.

GRADE EXPECTATIONS AS REFERENCE POINTS

You may remember taking college examinations and doing better or worse than expected. After a test, but

before grades are given, students often generate some grade expectation. After all, they know how much they

have studied and how well they know the material, and have some idea as to the difficulty of the instructor

and exam. This knowledge results in an intuitive assessment of what grade can be expected. An early study

by Murstein (1965) demonstrated that this ‘intuitive assessment’ is not always correct as some students are

better able to accurately predict their grades. He found that 80% of the students stated that they expected an

A or B, irrespective of their actual ability to achieve such a grade. More recent research in the educational

domain has highlighted the interplay of these grade expectations on instructor evaluations (Holmes, 1972;

Marlin & Gaynor, 1989; Owie, 1985; Tata, 1999), subsequent academic achievement (Vollmer, 1984), and

affect experienced as a result of the grading context (Goldstein & Strube, 1994). The extensive research by

Dweck and her colleagues (for a summary, see Dweck, 1986) also suggests that students’ goal orientations

(learning vs. performance) differentially affect their reactions to performance-related outcomes in the

educational domain.

376 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 3: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

PREVIOUS APPROACHES TO SATISFACTION AND FAIRNESS

In Adams’ (1965) original paper and in studies since (Austin, McGinn, & Susmilch, 1980; Austin & Walster,

1974; Messe & Watts, 1983), equity theory has posited that people who receive an equitable outcome are

significantly more satisfied or feel that the outcome is more fair than those who are either underrewarded or

overrewarded. Additionally, satisfaction and fairness evaluations both affect similar organizationally rele-

vant variables (e.g. turnover, organizational commitment, trust). As a result, researchers have sometimes pre-

sented or discussed satisfaction and fairness as one construct (e.g. Austin & Walster, 1974) or together in a

single scale (Gilliland, 1994; Greenberg, 1990a). On the other hand, McFarlin and Sweeney (1992) recog-

nized that satisfaction and fairness are different and found that fairness judgments are predictive of satisfac-

tion. However, van den Bos et al. (1998) caution that, ‘In recent years less and less attention has been paid to

potential differences between these two types of judgments’ (p. 1493).

Three recent studies (Ordonez, Connolly, & Coughlan, 2000; van den Bos et al., 1997, 1998) have docu-

mented the differential effects of reference points on judgments of outcome satisfaction and outcome fair-

ness, although the results are not entirely congruent with one another. These differences will be discussed in

the sections that follow on fairness and satisfaction.

When comparing theories of fairness and satisfaction, it is also possible to distinguish the antecedents of

each construct. In the following sections we discuss theories of fairness and satisfaction and generate hypoth-

eses regarding the role of expectations in each.

FAIRNESS

Equity theory (Adams, 1965) predicts that when one’s outcome-to-input ratios are larger than that of a refer-

ent other, a perceived overpayment inequity occurs and perceived fairness decreases (e.g. Austin, McGinn, &

Susmilch, 1980). In terms of exam grades, students should feel overpayment inequity if they have identical

inputs to referent students yet receive higher outcomes. Equity theory traditionally assumes that these com-

parisons are social comparisons made among people. However, as discussed earlier, it has also been sug-

gested that individuals often make comparisons of outcomes to self-created expectations instead of

referent others. An interesting question then develops: ‘When a participant views himself as the referent

other, does overpayment inequity (as proposed by equity theory) still occur?’

Self is often chosen as a referent because of the availability of relevant information and because

self-comparisons are ‘cognitively easier’ than social comparisons (see Kulik & Ambrose, 1992). In the

present study we include an individual’s expectation as a reference point from which outcomes can be

evaluated. These expectations appear to be central to the definition of fairness in the grading context as

some research has stated that grade fairness is simply a matter of ‘meeting expectations’ (cf. Tata, 1999).

This ‘met expectations’ approach is only instructive, to a point. What happens when the grade exceeds

the expectation? Following from equity theory, we propose that a fairness evaluation of an outcome is

based on matching the expectation and the outcome actually received—an expectation-matching

hypothesis. If the expected outcome and the actual outcome match, the outcome is judged as fair. If the actual

outcome is different (either positively or negatively) than the expectation, the outcome is considered unfair.

Therefore, fairness assessments are based on the congruency between expected outcome and actual outcome

received.

In the classroom setting, it is reasonable to expect that higher actual grades, regardless of expectations, are

going to be viewed as more fair than lower grades. In addition, consistent with the expectation matching

hypothesis, we expect that the congruence between expected grades and actual grades will impact fairness

evaluations. Therefore, we test the following hypothesis:

B. Cherry et al. Grade Expectations 377

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 4: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Hypothesis 1a: When expectations match actual outcomes, students will perceive the grade as being fairer

than when their grades fail to match the expectation.

In addition to the above hypothesis, fairness may depend on both the direction of and the extent to which

an outcome deviates from a reference point. There are two possible deviations that can occur relative to a

reference point; the discrepancy can either be positive (expect B, get A) or negative (expect A, get B). If the

discrepancy is positive, overreward occurs; one received more than she expected. With a negative discre-

pancy, underreward is present. Also, these discrepancies can have a magnitude of change associated with

each. For instance, a negative difference of ‘expect A, get B’ can be viewed as a one-level magnitude change,

while ‘expect A, get D’ can be viewed as a three-level magnitude change. Greenberg (1988) found that a

participant’s fairness perceptions were in proportion to the ‘magnitude of the inequity experienced.’ His find-

ings support equity theory’s prediction (Adams, 1965, p. 281) of a relationship between fairness and the

magnitude of the inequity—the greater the inequity, the greater the unfairness.

Past research in this area has suggested, however, that not all inequity is experienced in the same manner.

For example, Greenberg (1988) found that reactions to underrewarded situations (or ‘underpayment’)

were more pronounced than reactions to situations of overreward (or ‘overpayment’). Other studies (e.g.

Loewenstein, Thompson, & Bazerman, 1989; Messick & Sentis, 1985; Wicker & Bushweiler, 1970) have

found similar results suggesting that reaction thresholds for overpayment and underpayment inequity are

different. In general, these studies point out that individuals prefer equity when comparing outcomes to a

reference point (e.g. referent other). However, not surprisingly, if equitable outcomes are not possible, indi-

viduals prefer advantageous inequity to disadvantageous inequity. While this ‘egoism hypothesis’ (Wicker &

Bushweiler, 1970) is well established for external reference points, including multiple external reference

points (Ordonez, Connolly, & Coughlan, 2000), it has not yet been thoroughly examined with self-generated

expectations as internal reference points. An interesting study by van den Bos and colleagues (1998) alter-

natively proposed that there would be no evidence of egoism with regards to fairness judgments when using

either social-comparison based reference points or expectations. Although their results do not support the

egoism hypothesis for fairness judgments based on social comparisons, it would be difficult to make a con-

clusion based on expectations. This is because they only had two outcome-expectation conditions (‘better

than expected’ and ‘worse than expected’). An ‘equal to expectation’ condition was not included.

Given the prior research and the egoism hypothesis, we predict that when participants experience advan-

tageous inequity (receive a higher grade than expected) they will reevaluate their inputs (‘I guess I knew

more than I thought’) and perceive the situation as more fair than when they receive disadvantageous

inequity (receive a lower grade than expected). Therefore, we test the following egoism hypothesis:

Hypothesis 1b: When expectations match actual outcomes, students will perceive the grade as being fairer

than when their grades fail to match or exceed the expectation. However, failing to meet expectations will

be perceived as less fair than exceeding expectations.

SATISFACTION

Many satisfaction theories are based on comparative information and ‘the notion that individuals make per-

formance judgments with reference to a standard is not unique to any one field, and no one discipline can lay

claim to it’ (Oliver, 1997, p. 99). The question then becomes what is the correct reference point for satisfac-

tion in this context? Is it a self-generated expectation of performance that serves as the reference point for

satisfaction or is there a better reference point for satisfaction? Locke (1976) proposed that satisfaction is a

result of outcome–value congruence; outcomes that are congruent with established values will be viewed as

satisfying. Therefore, in Locke’s theory, the correct reference point is the participant’s established values

relative to the outcome at hand. Specifically, the satisfaction evaluation will be based on whether one obtains

378 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 5: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

what one values—a value-matching hypothesis. Locke defined a value as ‘what a person consciously or sub-

consciously desires, wants, or seeks to attain’ (1976, p. 1304). For many outcomes, such as money, security,

and achievement, more is preferred to less. Based on this line of reasoning, it can be assumed that students

value higher grades as opposed to lower grades (an ‘A’ is more favorable than a ‘D’) and, thus, are more

satisfied with higher grades.1 Locke reasoned that ‘a promotion or rise will ordinarily give pleasure, whether

it is expected or not, while a demotion or reduction in salary will be unpleasant, whether it is expected or not’

(1976, p. 1303). This suggests that expectations play a less-important role in satisfaction with an outcome

than they do in fairness perceptions.

Values can be considered as more enduring than expectations. Similarly, values are relatively constant,

while expectations can fluctuate depending on the situation. If satisfaction is driven by the congruence

between what we obtain and what we value, then expectations will play less of a role in influencing satisfac-

tion evaluations. Therefore, we propose that satisfaction is determined through a value-matching proposition

whereby if outcomes match what you value (not what you expected), you will be satisfied. Given our

assumption that students value high grades more than low grades, we reasoned that:

Hypothesis 2: Actual grades, rather than expectations, will be the primary determinant of satisfaction.

Hypothesis 2 may be seen as conflicting with the Expectancy Disconfirmation Model of Satisfaction

(Oliver, 1980; for a review, see Yi, 1990) because our hypothesis rests on the actual outcome being more

important than the expectation. Oliver (1997) discusses the superiority of expectations as the appropriate

reference point for evaluations of product satisfaction, but he also acknowledges that the actual outcome

or performance of the product can dominate the satisfaction response. Beyond performance expectations,

other researchers (e.g. Spreng, MacKenzie, & Olshavsky, 1996) have suggested that satisfaction is partially

shaped by the extent to which the product or service fulfills an individual’s desires. These desires are defined

by Spreng and colleagues as ‘the levels of attributes and benefits that a consumer believes will lead to . . .higher-level values’ (1996, p. 17).

Oliver (1997) acknowledges that exceeding expectations is always better (in terms of satisfaction) than

just meeting expectations. Much research in the satisfaction domain suggests that this positive disconfirma-

tion (receiving more than one expected) is accompanied by an increased satisfaction evaluation. However, it

is also widely known that positive disconfirmations are not experienced in the same manner as negative dis-

confirmations (receiving less than one expected). In fact, prospect theory’s (Kahneman & Tversky, 1979)

loss aversion would suggest that receiving less than one expected would see a significantly greater decrease

in satisfaction than would receiving more than one expected. Ordonez, Connolly, and Coughlan (2000) found

that satisfaction ratings of salary offers relative to a single comparison other displayed this loss-aversion

pattern. In light of this, we offer the following hypothesis:

Hypothesis 3: Higher grades will be rated as more satisfying but will be asymmetrically related to expec-

tations. That is, one grade lower than expected will decrease satisfaction more than one grade higher than

expected will increase satisfaction.

In a series of three studies, we examined the impact of grade expectations on fairness and satisfaction

judgments of exam grades in a college setting. Study 1 employs a fully controlled experimental scenario

design and Studies 2 and 3 extend the investigation to a naturalistic field survey of real-grade expectations

and actual exam grades received by students.

1The assumption that is made here regarding the value that students attach to grades is based on the research of Milton, Pollio, and Eison(1986). The results from their National Grade Survey, which included more than 6000 respondents, suggested that students not onlyvalue higher grades, but they actually overvalue As and Bs.

B. Cherry et al. Grade Expectations 379

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 6: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

STUDY 1: EXPERIMENTAL DESIGN AND LABORATORY SETTING

Method

Participants were presented with a set of hypothetical scenarios. In each scenario, an actual exam grade and

an expected exam grade were presented. Participants judged their satisfaction with their performance and the

fairness of the exam. The title of this study indicates that the stimuli were experimentally manipulated and

the participants made their responses in a laboratory setting.

Participants

Sixty-eight upper division undergraduate business students at a large state university received course credit

for their participation. The mean age was 24.7 years and a majority (63%) was male. Mean self-reported

GPA was 3.12 on a 4.0 scale.

Procedures

Participants read a brief introduction that was designed to engage them in the role of a student who had just

received the results of an exam. The introduction encouraged them to recall past expectations they had about

an exam grade. Because the sample only included current students, we believe it was not difficult for them to

imagine themselves as students who have just received exam grades. Specifically, the participants were

instructed: ‘In the following eighteen scenarios, imagine you are the student who expected to receive a grade

and received an actual grade that may or may not be the same as the expected grade. How would you feel?’

After each expected grade and actual grade combination was presented, participants rated: (1) their opinion

(as that student) about the fairness of the exam; and (2) their satisfaction with their performance on the exam.

Both ratings were on seven-point scales, anchored at 1¼ very unfair to 7¼ very fair, and 1¼ very unsatisfied

to 7¼ very satisfied. After completing the eighteen scenarios, participants completed a brief demographic

questionnaire (age, GPA, gender).

Materials

All participants were given stimulus packets that contained 18 randomized combinations of expected and

actual grades with each scenario followed by a fairness and satisfaction measure. Sixteen scenarios were

created from an actual grade (4)� expected grade (4) within-subjects, factorial design. Levels for both actual

and expected grades were A, B, C, and D. The 16 scenarios were randomized into four different stimuli

orders. To assess respondents’ consistency across scenarios, two random scenarios were replicated. Fairness

and satisfaction ratings for the repeated scenarios were consistent across the first and second responses

(r¼ 0.78 for fairness and r¼ 0.96 for satisfaction), indicating that participants’ responses to both measures

were reliable. The ratings for these replicated items were not used in the analyses that follow.

ResultsMean fairness and satisfaction ratings are plotted in Figure 1, panels A and B, respectively. Ratings are

plotted as a function of expected grade, with a separate curve for each level of the actual grade received.

Note that both expected and actual grade labels are subscripted with the letter ‘H’ for hypothetical. In Studies

2 and 3, the subscript ‘R’ for real will be used to denote expected and actual grades experienced by the

participants.

A 4 (actual grade)� 4 (expected grade)� 4 (order)� 2 (response dimension: fairness vs. satisfaction)

MANOVA was conducted, where the order factor was between subjects and all other factors were within

380 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 7: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

subjects. There are no order main effects for either fairness or satisfaction and there are no significant inter-

actions with order for fairness. There are, however, two significant interactions with order for satisfaction:

(1) expected grade� order (F9, 186¼ 2.85; p¼ 0.004); and (2) expected grade� actual grade� order

(F27, 558¼ 2.85; p¼ 0.024). Plots of means were examined to investigate the source of the order effects.

However, no clear discernible pattern was determined. The order effects were due to one of the four versions.

All analyses for Study 1 were conducted with and without this version. The general pattern of results was the

same, and only minor differences in the magnitude of the statistics were noted. Thus, the remaining analyses

will discuss the mean fairness and satisfaction ratings collapsed over order. The details of this analysis will

be discussed in the following paragraphs.

Fairness ratings

As both Hypotheses 1a and 1b imply, there should be an interaction between actual and expected grades for

fairness ratings. Examining Figure 1A, we can see that the mean fairness ratings are consistent with this

prediction. This actual� expected grade interaction is significant (F9, 558¼ 33.6; p< 0.001) as well as both

main effects for actual grade (F3, 186 ¼ 89.83; p< 0.001) and expected grade (F3, 186¼ 6.73; p< 0.001).

However, Hypotheses 1a and 1b make finer-grained predictions that cannot be properly examined with

the overall MANOVA. In order to fully test these predictions, the data need to be examined in terms of devia-

tions from the expected grade.2

The mean fairness and satisfaction ratings are replotted in Figure 2 (panels A and B, respectively) in terms of

deviations from expected grade. The actual–expected difference is either negative, zero, or positive to indicate

that the student received a lower, same, or higher grade than expected, respectively. The value of the number

denotes the size of the difference between actual and expected grades. For example, an actual–expected value

Figure 1. Mean fairness and satisfaction ratings from Study 1 as a function of expected grade. The subscript ‘H’ indicatesthat the actual and expected grades were hypothetical

2Comparing the average ratings relative to deviations from expected grades appears, on the surface, to suffer from the problems inherentin difference scores. However, note that the ratings themselves are not difference scores, rather the independent variable is a deviationbetween actual and expected grades. Thus, the problems that have been revealed with difference scores (Edwards, 1995) should notapply here.

B. Cherry et al. Grade Expectations 381

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 8: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

of �1 can represent three different combinations (actual, expected): (B, A), (C, B), and (D, C). Similarly, a

value of 2 can represent (A, C) or (B, D). Consistent with previous literature, positive and negative differences

will be denoted ‘advantageous inequity’ and ‘disadvantageous inequity,’ respectively.

Consistent with our Hypothesis 1a, Figure 2A shows that mean fairness ratings increase with actual grade

and any deviations between actual and expected grades (both advantageous and disadvantageous inequity)

decrease fairness ratings. However, the fairness ratings do not appear to follow the egoism hypothesis since

the slopes on either side of the expectation are approximately the same size. Contrary to Hypothesis 1b,

receiving one grade higher than expected is rated to be as unfair as receiving one grade lower than expected.

To model the effects described in Figure 2A, a regression analysis was conducted that predicted fairness

ratings from actual grade and the actual–expected differences. Even though we did not find evidence of the

differential effects of disadvantageous and advantageous inequity on fairness we created two difference

terms in the regression: positive actual–expected (pos act–exp) and negative actual–expected (neg act–

exp). By creating these two difference terms, we were able to inspect the standardized regression weights

(�s) for each to test for asymmetric effects of differences from expectations. The positive (negative) differ-

ence term indicates scenarios in which the actual grade was above (below) the expectation. Note that the

values can range from 1 to 3, indicating the size of the absolute difference between actual and expected

grades. Coefficients and R2 values are presented in Table 1 for both fairness and satisfaction ratings along

with the results from Studies 2 and 3, which will be discussed later. Consistent with the effects seen in

Figure 2A, fairness increases with actual grade (�¼ 0.47, p< 0.001), and decreases with both positive

actual–expected (�¼�0.32, p< 0.001) and negative actual–expected (�¼ � 0.29; p< 0.001),3 thus,

Hypothesis 1a is supported. In addition, the similarity in the size of the difference term coefficients suggests

the egoism hypothesis (Hypothesis 1b) is not supported by these data.

Figure 2. Mean fairness and satisfaction ratings from Study 1 as a function of actual grade and actual–expected gradedifference. The subscript ‘H’ indicates that the actual and expected grades were hypothetical

3In addition, regressions were conducted separately for each participant. Of those who have significant (p< 0.05) difference coefficients,90% (44 of 49) are consistent with the aggregate results. That is, one or both difference coefficients are significantly negative. Each of theremaining 5 participants with significant difference coefficients has one difference coefficient significantly positive and the othersignificantly negative.

382 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 9: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Satisfaction ratings

In Figure 1B, mean satisfaction ratings are plotted as a function of expected grade. Hypothesis 3 implies that

actual and expected grades will interact for satisfaction ratings; visual inspection of Figure 1B supports this

prediction. The MANOVA results for satisfaction ratings indicate that the actual� expected grade interac-

tion is indeed significant (F9, 558¼ 29.9; p< 0.001) in addition to significant main effects for actual grade

(F3, 186¼ 899.1; p< 0.001) and expected grade (F3, 186 ¼ 88.1; p< 0.001). Supporting Hypothesis 2, the

effect size of actual grade ð�2p ¼ 0:935Þ is much higher than either those for expected grades

ð�2p ¼ 0:587Þ or the actual� expected grade interaction ð�2

p ¼ 0:326Þ.In order to fully examine Hypothesis 3 predictions, we must examine how satisfaction ratings vary as a

function of the actual–expected difference. Thus, Figure 2B shows mean satisfaction ratings as a function of

actual–expected grades. As with fairness, satisfaction increases with actual grade. Additionally, like fairness,

satisfaction decreases with negative actual–expected differences: the lower a grade is compared to the expec-

tation, the lower the satisfaction. Loss aversion can be seen in the fact that negative differences are more

dissatisfying than positive differences of the same magnitude are satisfying. Thus, it appears that expected

grades have an effect on satisfaction ratings, contrary to Locke’s (1976) hypothesis but in keeping with

results found in the consumer satisfaction literature (cf. Szymanski & Henard, 2001; Yi, 1990). This result

is also consistent with our Hypothesis 3. However positive actual–expected differences do not appear to

greatly influence satisfaction ratings.

A regression model was used to fit the satisfaction ratings and the results are shown in the lower half of

Table 1. Consistent with Figure 2B, increases in actual grade significantly increased satisfaction (�¼ 0.61;

p< 0.001). The fact that the negative actual–expected coefficient (�¼�0.29; p< 0.001) is negative and lar-

ger than the positive coefficient for positive actual–expected (�¼ 0.09; p< 0.001) indicates that lower

Table 1. Multiple regression results from studies 1 to 3 for both fairness and satisfaction ratings

Fairness

Coefficients (�) Study 1 (Exp.a) Study 2 (Exp. Exama) Study 3 (Exam)

Actual grade 0.474* 0.626* 0.388*Pos act–exp �0.320* �0.202* �0.020Neg act–exp �0.287* �0.187* �0.360*R2 0.493 0.724 0.459F 14.14 10.30 24.04(d.f. num, d.f. denom) (70, 1017) (168, 829) (3, 85)

Satisfaction

Coefficients (�) Study 1 (Lab.a) Study 2 (Lab. Exama) Study 3 (Exam)

Actual grade 0.614* 0.651* 0.649*Pos act–exp 0.087* �0.001 �0.101Neg act–exp �0.289* �0.313* 0.124R2 0.800 0.894 0.608F 57.90 33.09 43.93(d.f. num, d.f. denom) (70, 1015) (168, 829) (3, 85)

aDue to the within-subjects design, individual participant dummy variables were added to the regression analysis to account for therepeated measures effect. Individual participant coefficients are not shown.*p< 0.001.

B. Cherry et al. Grade Expectations 383

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 10: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

grades relative to the expectation decrease satisfaction more so than higher grades increase satisfaction (sup-

porting Hypothesis 3).4

Discussion

As mentioned earlier, some researchers have treated fairness and satisfaction as one construct (cf. Austin &

Walster, 1974; Gilliland, 1994; Greenberg, 1990a). Indeed, fairness and satisfaction are related (McFarlin &

Sweeney, 1992; Sweeney & McFarlin, 1993) as we found that the two items are highly correlated (r¼ 0.54).

As evidenced from Figure 2, fairness and satisfaction are two different constructs that show distinctly dif-

ferent relationships to actual–expected grade differences. The regression analysis coefficients for the differ-

ence terms also indicate that the ratings differ. Grades higher or lower than the expected grade are judged to

be unfair relative to those that met expectations. Thus, the fairness ratings are consistent with equity theory

predictions that both overpayment (advantageous inequity) and underpayment (disadvantageous inequity)

create feelings of unfairness. This is not the pattern found with satisfaction ratings—while grades lower than

the expected grade are dissatisfying, grades higher than expected increase (not decrease) satisfaction. The

satisfaction ratings are consistent with prospect theory (Kahneman & Tversky, 1979) which suggests that

‘losses loom larger than gains.’ The subjective utility associated with a grade higher than one expects is posi-

tive, but not nearly as strong as the negative prospect of receiving less than one expects. These results parallel

those of Ordonez, Connolly, and Coughlan (2000) who examined satisfaction ratings to hypothetical salary

offers when compared to one or more social referents. For example, participants were more unsatisfied by

receiving an offer $10 000 less than a comparison other than they were satisfied by receiving $10 000 more

than another comparableperson.

The design of Study 1 provides a controlled examination of perceptions of fairness and satisfaction for a

situation that is very familiar to our student participants. However, there is a trade-off between experimental

control and obtaining realistic reactions in natural settings. Study 1 has a few flaws that may impede the

generalizability of the results. For example, due to the within-subjects design, demand characteristics

may have been present. In addition, student participants were presented with only two pieces of information

in the hypothetical scenarios—the expected grade and the actual grade. In a sense, the first study asks par-

ticipants to evaluate outcomes devoid of context. It could be expected that in natural settings, individuals

utilize many pieces of information when evaluating an outcome. Study 2 was designed to maintain some

experimental control but also to include reactions in naturally representative situations. This study used

the hypothetical grade stimuli presented in Study 1 in a realistic environment—an actual course examina-

tion. The final study (Study 3, described later) addresses the within-subject limitation and will obtain

actual fairness and satisfaction judgments from students who personally experience an expected versus

actual grade discrepancy. Results from all three studies will give us a more complete picture of how these

judgments are made.

STUDY 2: EXPERIMENTAL DESIGN AND A CLASSROOM SETTING

Study 2 is a lab/field study hybrid, since grades from the hypothetical stimuli from Study 1 are used in con-

junction with grade expectations from an actual course exam. Immediately following the completion of an

4Again, separate regressions were conducted for each participant. Of those who have significant (p< 0.05) difference coefficients, 90%(35 of 38) are consistent with the aggregate results. That is, one or both difference coefficients are significant in the same direction as inthe aggregate (i.e. a positive coefficient for the positive difference and a negative coefficient for the negative difference). Each of theremaining three participants with significant difference coefficients has a significantly negative coefficient for the positive differenceterm and a nonsignificant coefficient for the negative difference term.

384 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 11: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

exam, participants were asked to rate their feelings if they later learn that they received an A, B, C, D, or F

grade. They were then asked to state their personal expected grade. Thus, the expected grades were real and

derived by the student, while the actual grades were hypothetical. We chose this type of design because we

knew that expected grades and actual grades would be highly correlated and would thus limit an equal dis-

tribution of participants in the cells of the expected grade by actual grade matrix. We recognized that it would

be difficult to obtain adequate cell sizes in the off-diagonal cells. This design ensured that a full range of

actual grades and expected grades would be included in the analysis.

Method

Participants

One hundred and seventy-two upper-division undergraduate business students, enrolled in the same class at a

large state university, participated in the experiment. Six of the participants did not complete the survey and

were excluded from the analyses, leaving a sample size of n¼ 166. The mean age for participants was 22.7

years, with a slight majority (52%) of female students. The mean self-reported GPA was 3.16 on a 4.0 scale.

Procedures

Student participants responded to a self-report questionnaire immediately after completing a regularly

scheduled course exam. An experimenter, who was not the course instructor, administered all questionnaires.

Students were told that the instructor would not receive any of the information given in the questionnaire

(except in aggregate form). Students were not required to participate, however all students volunteered.

Measures

One half of the participants were presented with the letter grades A, B, C, D, and F and, for each grade, were

asked to rate how fair the test would be if they were to receive this grade using a 1 (very unfair) to 7 (very

fair) scale. These participants were then presented with letter grades A, B, C, D, and F and were asked to rate

how satisfied they would be if they received each grade using a 1 (very dissatisfied) to 7 (very satisfied) scale.

The other half of the participants answered the satisfaction questions before the fairness questions. The five

letter grades were presented in a single matrix for both types of ratings; thus, participants rated the five

grades virtually simultaneously. Due to this format, we did not vary the order of the letter grades to test

for order effects. After completing these fairness and satisfaction questions, participants stated the letter

grade they expected to receive on the exam they had just finished. Finally, they answered general demo-

graphic questions (age, GPA, gender).

Results

Consistent with Murstein’s (1965) results, our students were overly optimistic about their grade expecta-

tions: 24 expected an A; 102 expected a B; 38 expected a C; 2 expected a D; and none expected an F. Given

the lower cell sizes for low expected grades, participants expecting either a C or a D were collapsed into a

single group for the means presented in Figures 3 and 4 and the corresponding MANOVA. A MANOVA was

conducted on the rating responses in a 5 (actual grade: A, B, C, D, F)� 3 (expected grade: A, B, C–D)� 2

(order: fairness 1st, satisfaction 1st)� 2 (response dimension: fairness, satisfaction) design, with both actual

grade and the response dimension as within-subjects and expected grade and order as between-subjects.

There are no significant order main effects or interactions of order with the other factors for either fairness

or satisfaction ratings. Therefore, the following analyses collapse over order and will be discussed in the

following paragraphs.

B. Cherry et al. Grade Expectations 385

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 12: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Fairness ratings

As mentioned previously, Hypotheses 1a and 1b imply that fairness ratings should reveal an actual� expected

grade interaction. Both mean fairness and satisfaction ratings are plotted against expected grade in Figure 3

(Panels A and B, respectively) in the same manner as Figure 1. The fairness means in Figure 3A have a similar

pattern as those in Figure 1A for Study 1—mean fairness ratings increase with actual grade and actual and

expected grades appear to interact. As expected, the MANOVA reveals a significant actual grade� expected

Figure 3. Mean fairness and satisfaction ratings from Study 2 as a function of expected grade. The actual grades aresubscripted with ‘H’ to signify that they were hypothetical values, whereas expected grades are subscripted with ‘R’

to indicate that they were real grades expected by the participants

Figure 4. Mean fairness and satisfaction ratings from Study 2 as a function of actual grade and actual–expected gradedifference. The actual grades are subscripted with ‘H’ to signify that they were hypothetical values, whereas expected

grades are subscripted with ‘R’ to indicate that they were real grades expected by the participants

386 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 13: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

grade interaction (F8, 520¼ 8.94; p< 0.001) and main effect for fairness ratings of actual grade (F4, 652¼ 206.8;

p< 0.001); however, expected grade (F2, 163¼ 0.67; p¼ 0.513) was not significant. While these results are

consistent with Hypotheses 1a and 1b, the actual predicted pattern cannot be investigated by examining the

data in this format. Thus, as with Study 1, the fairness ratings will be examined in relation to actual–expect

grade differences in a plot of the means and multiple regression analysis.

Figure 4 plots the mean fairness (Panel A) and satisfaction (Panel B) ratings against the actual–expected

grade difference, with a separate curve for each level of actual grade, similar to Figure 2. The patterns in

Figure 4A closely resemble those in Figure 2A, suggesting that fairness results are similar whether con-

ducted in the laboratory with a purely hypothetical scenario or in an actual examination setting with real

expected grades. As in Study 1 (consistent with Hypothesis 1a), for a given level of actual grade, fairness

is highest when actual grades are the same as expected grades. Fairness decreases when grades are higher or

lower than expected. Also, as with Study 1, the means do not provide support for the egoism prediction for

Hypotheses 1b—the positive and negative differences appear to decrease fairness roughly equally. Finally,

fairness generally increases with increases in actual grade.

The same regression model used in Study 1 was fitted to fairness and satisfaction ratings in Study 2 (see

Table 1). For fairness, the actual grade had a significant positive effect (�¼ 0.63; p< 0.001). Both positive

(�¼�0.20; p< 0.001) and negative actual–expected difference (�¼�0.19; p< 0.001) terms have signifi-

cant negative coefficients, suggesting that grades different from those expected are less fair than those that

met expectations (supporting Hypothesis 1a). Also, because the difference coefficients are similar, there is no

evidence to support the egoism hypothesis (Hypothesis 1b).

Satisfaction ratings

Mean satisfaction ratings plotted against expected grade in Figure 3B have the same general pattern as in

Figure 1B. Satisfaction increases with increasing actual grade and displays an actual grade� expected grade

interaction. While the main effects for actual grade (F8, 652¼ 1267.4; p< 0.001) and expected grade

(F2, 163¼ 11.5; p< 0.001) are both significant, the effect size for actual grade ð�2p ¼ 0:89Þ is much higher

than those for either expected grade ð�2p ¼ 0:12Þ or the actual� expected grade interaction ð�2

p ¼ 0:10Þ;which is consistent with Hypothesis 2. The actual grade� expected grade interaction (F8, 652¼ 8.7;

p< 0.001) is significant, consistent with Hypothesis 3.

However, to fully test the predictions for Hypotheses 2 and 3, the satisfaction ratings need to be compared

relative to the actual–expected grade differences. Mean satisfaction ratings plotted against actual–expected

grade differences in Figure 4B mirror those in Study 1. Consistent with Hypothesis 2, there is a strong main

effect of actual grade. In addition, Hypothesis 3 is supported—negative differences from expectations

decrease satisfaction more than positive differences increase satisfaction. Satisfaction decreases with nega-

tive actual–expected grade differences, suggesting that receiving a grade lower than expected is less satisfy-

ing. Positive actual–expected grade differences do not appear to influence satisfaction.

For the regression analysis on the satisfaction ratings (Table 1), the actual grade has a significant, positive

effect (�¼ 0.65; p< 0.001). The fact that the beta coefficient for actual grade is much larger than both the

difference terms provides support for Hypothesis 2. The negative difference term has a significant negative

effect (�¼�0.31; p< 0.001), whereas the positive difference term is not significant (�¼�0.001; p¼ 0.94).

This suggests the same asymmetric effects found in Study 1 where negative differences have a stronger

impact on satisfaction ratings than positive differences (supporting Hypothesis 3).

Discussion

The main conclusion to be drawn from Study 2 is that results are similar whether conducted in a controlled

laboratory setting or in a real examination setting. Advantageous inequity was seen as equally unfair as

B. Cherry et al. Grade Expectations 387

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 14: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

disadvantageous inequity, but the former was judged more satisfying than the latter. One slight difference

between Studies 1 and 2 occurred with the loss aversion effect. In both studies, the negative actual–expected

difference had a significant negative impact on satisfaction. The positive difference produced a significant

positive effect on satisfaction ratings in Study 1 but no significant effect in Study 2. This may be due to the

fact that Study 2 used expected grades that were based on the participants’ predictions of their own perfor-

mance. Since most participants (77%) expected to receive an A or B, there were fewer observations above the

expected grade (i.e. positive actual–expected values) than below or at the expected grade. This makes it dif-

ficult to find an effect of positive differences in Study 2.

Despite the similarities between Studies 1 and 2, the same criticisms of Study 1 can be made of Study 2—

the ‘actual’ grades examined were, in fact, hypothetical values. While there is merit in studying reactions to

hypothetical stimuli, we would be remiss to conclude our study with only these scenario-based responses.

Study 3 addresses this issue by using actual and expected grades experienced by in-class student participants.

It is interesting to see if students who receive a grade higher than they expected report that the test is unfair as

students did in the previous studies.

STUDY 3: CLASSROOM SETTING

In this study, we measure participants’ expectations immediately before receiving their actual grades and mak-

ing their fairness and satisfaction judgments. The positive aspect of this study is that we are measuring actual

feelings in a realistic, natural situation. Other researchers (e.g. Gilliland, 1994; Greenberg, 1988, 1990a) have

conducted studies in this manner. While realism is increased, the study’s negative aspect is the lack of control

over many variables that affect fairness and satisfaction judgments other than expected and actual grades, such

as the procedural justice of the examination process. In addition, since actual and expected grades will

undoubtedly be positively correlated, it will not be possible to achieve a full factorial design.

Method

Participants

Eighty-nine upper-division undergraduate business students at a large state university participated in the

experiment during a scheduled class session. Due to a time constraint, no demographic data were collected.

Procedures and measures

Participants met in small, weekly discussion sections (about 20–30 students). Prior to receiving their exam

grade, they were asked to state the grade they expected to receive (A–F). Then they were shown their actual

exam grades. On the second page of the questionnaire, participants then indicated the grade they actually

received (A–F). Participants rated the fairness of the grade they received with a Likert-type scale (1¼ very

unfair to 7¼ very fair). Participants also rated their satisfaction with their performance (1¼ very unsatisfied

to 7¼ very satisfied).

Results

As expected, several of the cells in the actual� expected grade matrix were either small or empty; for exam-

ple, no F grades were given or expected and few D grades were given or expected. In order to have sufficient

data points in the cells, both actual grade and expected grade were collapsed to have three levels: A, B, and

C–D. These collapsed cells were used in the MANOVA and to display the mean ratings in Figures 5 and 6.

Data were not collapsed in the regression analyses that are described below. A MANOVA was conducted on

388 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 15: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

the rating responses in a 3 (actual grade: A, B, C–D)� 3 (expected grade: A, B, C–D)� 2 (rating dimension:

fairness, satisfaction) design, with the expected and actual grade factors as between-subjects and the

response dimension as within-subjects.

Fairness ratings

Figure 5 presents mean ratings for fairness and satisfaction as a function of expected grade (Panels A and B,

respectively). Fairness means increase with actual grade; the main effect is significant (F2, 80¼ 17.67;

p< 0.001). However, fairness is only weakly related to expected grade. The main effect for expected grade

Figure 5. Mean fairness and satisfaction ratings from Study 2 as a function of expected grade

Figure 6. Mean fairness and satisfaction ratings from Study 3 as a function of actual grade and actual–expected gradedifference. The subscript ‘R’ indicates that the actual and expected grades were real (experienced by the participant)

B. Cherry et al. Grade Expectations 389

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 16: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

is only marginally significant (F2, 80¼ 2.071; p¼ 0.07) and the actual� expected grade interaction is not

significant (F4, 80 ¼ 1.68; p¼ 0.16). Thus, it appears that Hypothesis 1a is supported but Hypothesis 1b is

only weakly supported. In order to fully test the predictions of both Hypotheses 1a and 1b, fairness ratings

will be examined relative to actual–expected grade differences.

Figure 6 presents mean ratings plotted for the fairness and satisfaction ratings (presented in Panels A and

B, respectively). Due to the small sample sizes in the actual� expected grade matrix, the actual–expected

difference was collapsed into actual grade below, equal to, or above expected grade. Mean fairness ratings

increased with increases in actual grade as did the fairness ratings in Studies 1 and 2. However, fairness

ratings differ from Studies 1 and 2 on the actual–expected difference. In Study 3, grades lower than expected

decrease fairness ratings. However, grades higher than expected grades are judged to be as fair as grades

equal to the expectation. Recall that in Studies 1 and 2, participants reported that grades higher than expected

were less fair than those that met expectations.

The fairness and satisfaction ratings were fitted to the same regression model used in Studies 1 and 2—

coefficients and fits are presented in Table 1. For fairness, actual grade has a significant positive effect

(�¼ 0.39; p¼ 0.001). The negative actual–expected difference coefficient is significantly negative

(�¼�0.36; p¼ 0.001), but the positive actual–expected difference is not significant (�¼�0.02;

p¼ 0.82). Thus, there is an effect of expectations but in a somewhat different pattern than in the scenario

studies, since grades above the expected grade are not rated as less fair than grades that met expectations.

Satisfaction ratings

Mean satisfaction ratings are plotted against expected grade in Figure 5B. Actual grade has the most obvious

impact on satisfaction ratings which yields a significant main effect (F2, 80 ¼ 36.4; p< 0.001). While the

slopes of the curves suggest an effect of expected grade, this main effect (F2, 80¼ 2.23; p¼ 0.11) is not sig-

nificant, nor is the actual� expected grade interaction (F4, 80¼ 0.55; p¼ 0.70). Thus, it appears that Hypoth-

esis 2 is supported but there is no evidence supporting Hypothesis 3.

Examining the mean satisfaction ratings relative to the actual–expected grade differences (Figure 6B),

satisfaction increases with actual grade and the actual–expected grade difference as in Studies 1 and 2.

Unlike the satisfaction ratings in the previous two studies, there is no evidence of asymmetric effects from

the reference point—the slopes of the lines from below to equal are approximately the same size as those

from equal to above. The regression results in the lower half of Table 1 confirm these patterns. The actual

grade has a significant positive effect (�¼ 0.65; p< 0.001). However, neither the positive difference

(�¼ 0.124; p¼ 0.10) nor the negative difference terms (�¼�0.101, p¼ 0.28) are significant. Thus, the dif-

ference between expected and actual grades did not impact satisfaction ratings overall or asymmetrically.

We, therefore, find support for Hypothesis 2 but not Hypothesis 3. In sum, both actual grade and the

actual–expected difference had an effect on satisfaction in the scenario-based studies (Studies 1 and 2),

whereas only actual grade significantly impacted satisfaction in the field (Study 3).

DiscussionThere appear to be differences between the field (Study 3) and the lab (Study 1) in the participants’ fairness

judgments. Comparing fairness ratings (Figures 2A and 6A) we see that the effect of deviations from an

expected grade is different in the two settings. In the lab, both receiving a grade lower and higher

than expected are judged as less fair than receiving a grade that meets the expectation. In the field setting

of Study 3, only grades lower than expected are seen as less fair. Participants who actually received a grade

higher than they expected reported their grade as equally fair as those who had received grades that met their

expectation.

390 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 17: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

In addition, there appear to be differences between satisfaction ratings gathered in the field (Study 3) and

the lab (Study 1). In both settings, actual grade had a large, positive impact on satisfaction. However, expec-

tations affected satisfaction in the lab, but did not in the more natural field study. These results of Study 3 are

consistent with Locke’s ideas and our Hypothesis 2 that outcome–value congruence, but not outcome–expec-

tation congruence, should affect satisfaction. These differences are mentioned in detail in the general dis-

cussion that follows.

GENERAL DISCUSSION

Two primary goals guided our research. Our main goal in these studies was to understand how grade expec-

tations influence an individual’s judgment of fairness and satisfaction with an exam grade. On a larger scale,

we were interested in determining the effect of an individual’s expectation on how the individual interprets a

given performance outcome. In particular, we wanted to know how expectations relate to perceptions of fair-

ness and satisfaction.

During the past half-century, much research has been directed towards understanding satisfaction as a con-

struct. From studies on job satisfaction (e.g. Locke, 1976) to investigations of consumer satisfaction (e.g.

Oliver, 1997), the domain has been significantly explored. The domain of fairness research has been directed

toward understanding fairness and how individuals interpret situations as fair or unfair. Although the satis-

faction and fairness lines of research can be considered related, little research has investigated the degree of

overlap between the two (with exception of Ordonez, Connolly, & Coughlan, 2000; van den Bos et al., 1998).

It was our intention in this paper to investigate a small portion of the apparent common ground by exploring

the influence of expectations on these two constructs.

Our second goal was to investigate the consistency of results across a controlled laboratory experiment

using hypothetical scenarios and a more highly externally valid field study using real students with real exam

results. While we recognize the inherent value in conducting controlled lab experiments, we also recognize

the shortcomings and limitations of such research. With a significant amount of research in the satisfaction

and fairness domain being conducted in an experimental setting, we wanted to determine the effectiveness of

a scenario-based experiment as compared to a field study.

The results from the three studies achieved these general goals. All of the studies support the assertion that

satisfaction and fairness are different constructs (e.g. Austin, McGinn, & Susmilch, 1980; Messe & Watts,

1983; van den Bos et al., 1998) and that expectations influence these two constructs very differently. Indi-

viduals give little consideration to their expectation (i.e. expected grade) when judging their satisfaction with

the exam grades. Instead, they appear primarily to use the outcome received (i.e. actual grade, as evidenced

by significance weights) as their basis for evaluation. The higher the grade they receive, the more satisfied

they are. This result has also been found in studies on pay satisfaction (e.g. Messe & Watts, 1983) and con-

sumer satisfaction (e.g. Page & Spreng, 2002; Tse & Wilton, 1988).

As Locke (1976) suggested, satisfaction is driven by a value-attaining proposition, in that individuals are

satisfied when they have received an outcome that they value and not necessarily an outcome that they

expect. This is not the case with fairness judgments. In evaluating the fairness of exam grades, individuals

not only use the actual outcome (i.e. actual grade) but also compare that outcome against an expected out-

come (i.e. expected grade). In all of the current studies, receiving less than one expects is considered less fair

than receiving the same as what one expects. However, the three studies demonstrated different results on

fairness judgments when an individual received more than expected. Interestingly, in contrast to our results,

van den Bos et al. (1998) found that expectations had a stronger impact on satisfaction than fairness.

In the laboratory with hypothetical scenarios, receiving more than expected is considered less fair than

receiving the same as expected. However, in the classroom setting, after receiving results from a real exam,

receiving more than expected is perceived to be just as fair as receiving the same as expected. Although we

B. Cherry et al. Grade Expectations 391

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 18: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

did not anticipate that fairness in the lab setting would be perceived as different from fairness in the field

study, we do believe that this is an interesting and useful finding. Perhaps one explanation comes from equity

theory itself. Past research in equity theory has found that the overpayment effect is an inconsistent result

(Mowday, 1996). In our study, it could be argued that the reason we achieved different results in the lab and

in the field is due to one of the inherent criticisms of equity theory—an individual’s given level of input is not

always known and may not be unidimensional. In our controlled scenario study, the only input in the O/I

equation was the grade expectation provided to the participants. In contrast, the rich context of the field stu-

dies may have allowed for multiple sources of input to factor into the fairness assessment. A student in the

field study may have been interpreting the fairness of the outcome based on their multiple inputs (e.g. amount

of time spent studying, test-taking disposition) instead of only their grade expectation.

The difference that was found between the lab and the field is not unique to this study. Other researchers

have also discovered that results from the controlled laboratory are not always in agreement with results

obtained in more natural field studies when investigating, for example, components of organizational justice

(cf. Cohen-Charash & Spector, 2001) or satisfaction with leaders (Tyler & Caine, 1981).

The differences between the fairness judgments in the lab versus the field could be interpreted by under-

standing the terms of the relationships between individuals operating within the testing environment.

Loewenstein, Thompson, and Bazerman (1989) found that participants were most satisfied when outcomes

were distributed equally among friends and cooperating individuals. However, they found that when com-

petition (i.e. a negative interpersonal relationship with other party) was introduced into the inequity context,

‘participants displayed a selfish shift—they became more concerned with their own payoff, independent of

the other party’s, and more tolerant of advantageous inequality’ (1989, p. 438). Possibly the participants in

the lab study interpreted the hypothetical scenarios as noncompetitive because other participant information

was unknown, whereas those participants in the field study experienced real competition with other students

(only disadvantageous inequality is unfair). The contrasting results of overpayment inequity found in ours

and other’s studies echo Greenberg’s (1990b) concern regarding the importance of conducting fairness or

justice research outside of the controlled laboratory setting.

Limitations and directions for future research

We have focused broadly on expectations in this study. In future research, this construct could be defined

more carefully. It may be that a person’s expectation also includes a bit of ‘deservedness’ and ‘hopefulness.’

It is probable that these types of ‘expectations’ may be related to individual difference variables (e.g. self-

confidence, self-esteem, self-efficacy) and thus influence fairness evaluations differently. For example,

deservedness (e.g. generalized feelings of respect, fair play) could involve procedural justice or interactional

justice concerns thereby influencing fairness judgments differently. Hopefulness, on the other hand, may

involve a less-strong expectation that could change quickly based on contrasting information. It is probable

that some students hoped for an ‘A’ but really thought they would receive a ‘C.’

Another limitation is that we did not measure what grades the students wanted or desired. In our value-

matching hypothesis for satisfaction, we made a rather strong assumption that students want to receive an A.

However logical that assumption may seem, future researchers may want to explore the outcomes that

recipients desire and want, and the strength of the value attached to a particular outcome or

evaluation. The consumer-satisfaction literature provides an appropriate framework for this type of investi-

gation into performance-related outcomes. In addition to expectations, research has shown that individuals

use desires as a basis or standard for evaluating satisfaction with products and services (e.g. Myers,

1991; Spreng, MacKenzie, & Olshavsky, 1996; Westbrook & Reilly, 1983). Future research should also

include the degree of confidence (high or low) that an individual has in their performance expectation, as

this has been found to moderate the effect of expectation disconfirmation on product satisfaction (Spreng

& Page, 2001).

392 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 19: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

An extension of this research could also focus on whether other forms of performance evaluations (e.g. job

performance appraisals) find the same result. It would be interesting to develop a study that investigates

responses to outcomes where distributions are constrained by a fixed quantity of resource. This could be

accomplished in a classroom situation where grades are distributed based on a bell curve or some fixed dis-

tribution (this was not the case in the current studies). This type of manipulation may help us understand

what happens with individuals’ expectations and fairness judgments when the individuals are in direct com-

petition for resources. In like manner, it would be interesting to investigate organizations that allocate per-

formance evaluations within a forced distribution to see if performance expectations affect fairness and

satisfaction perceptions.

The research on goal orientations (cf. Dweck, 1986) could also be used to further extend this research. In

particular, one might ask, ‘How do grade expectations vary depending on the student’s goal orientation?’

Similarly, do the results that were found in these three studies hold up across both learning-oriented students

and performance-oriented students. Although this was not included in the brief of the current study, the line

of research offers a number of interesting questions and could be included in future research.

REFERENCES

Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.), Advances in experimental social psychology(Vol. 2, pp. 267–299). New York: Academic Press.

Austin, W., McGinn, N. C., & Susmilch, C. (1980). Internal standards revisited: effects of social comparisons andexpectancies on judgments of fairness and satisfaction. Journal of Experimental Social Psychology, 16, 426–441.

Austin, W., & Walster, E. (1974). Reactions to confirmations and disconfirmations of expectancies of equity and inequity.Journal of Personality and Social Psychology, 30, 208–216.

Cohen-Charash, Y., & Spector, P. E. (2001). The role of justice in organizations: a meta-analysis. OrganizationalBehavior and Human Decision Processes, 86(2), 278–321.

Dweck, C. (1986). Motivational processes affecting learning. American Psychologist, 41, 1040–1048.Edwards, J. (1995). Alternatives to difference scores as dependent variables in the study of congruence in organizational

research. Organizational Behavior and Human Decision Processes, 64, 307–324.Gilliland, S. W. (1993). The perceived fairness of selection systems: an organizational justice perspective. Academy of

Management Review, 18, 694–734.Gilliland, S. W. (1994). Effects of procedural and distributive justice on reactions to a selection system. Journal of

Applied Psychology, 79, 691–701.Goldstein, M. D., & Strube, M. J. (1994). Independence revisited: the relationship between positive and negative affect in

a naturalistic setting. Personality and Social Psychology Bulletin, 20(1), 57–64.Greenberg, J. (1988). Equity and workplace status: a field experiment. Journal of Applied Psychology, 73, 606–613.Greenberg, J. (1990a). Employee theft as a reaction to underpayment inequity: the hidden cost of pay cuts. Journal of

Applied Psychology, 75, 561–568.Greenberg, J. (1990b). Organizational justice: yesterday, today, and tomorrow. Journal of Management, 16(2), 399–432.Holmes, D. S. (1972). Effects of grades and disconfirmed grade expectancies on students’ evaluations of their instructor.

Journal of Educational Psychology, 63(2), 130–133.Hom, P. W., Griffeth, R. W., Palich, L. E., & Bracker, J. S. (1999). Revising met expectations as a reason why realistic job

previews work. Personnel Psychology, 52, 97–112.Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under risk. Econometrica, 47, 263–291.Kulik, C. T., & Ambrose, M. L. (1992). Personal and situational determinants of referent choice. Academy of

Management Review, 17, 212–237.Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.), Handbook of industrial and

organizational psychology (Vol. 1, pp. 1297–1349). Chicago: Rand McNally.Loewenstein, G. F., Thompson, L., & Bazerman, M. H. (1989). Social utility and decision making in interpersonal

contexts. Journal of Personality and Social Psychology, 57(3), 426–441.Marlin, J. W., & Gaynor, P. E. (1989). Do anticipated grades affect student evaluations? A discriminant analysis

approach. College Student Journal, 23, 184–192.McFarlin, D. B., & Sweeney, P. D. (1992). Distributive and procedural justice as predictors of satisfaction with personal

and organizational outcomes. Academy of Management Journal, 35, 626–637.

B. Cherry et al. Grade Expectations 393

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 20: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Messe, L. A., & Watts, B. L. (1983). Complex nature of the sense of fairness: internal standards and social comparisonsas bases for reward evaluations. Journal of Personality and Social Psychology, 45, 84–93.

Messick, D. M., & Sentis, K. P. (1985). Estimating social and nonsocial utility functions from ordinal data. EuropeanJournal of Social Psychology, 15, 389–399.

Milton, O., Pollio, H. R., & Eison, J. A. (1986). Making sense of college grades: Why the grading system does not workand what can be done about it. San Francisco: Jossey-Bass.

Mowday, R. T. (1996). Equity theory predictions of behavior in organizations. In R. M. Steers, L. W. Porter, & G. A.Bigley (Eds.), Motivation and leadership at work (6th edn., pp. 53–71). New York: McGraw-Hill.

Murstein, B. I. (1965). The relationship of grade expectations and grades believed to be deserved to actual gradesreceived. The Journal of Experimental Education, 33(4), 357–362.

Myers, J. H. (1991). Measuring customer satisfaction: is meeting expectations enough? Marketing Research, 2, 35–43.Oliver, R. L. (1980). A cognitive model of the antecedents and consequences of satisfaction decisions. Journal ofMarketing Research, 17, 460–469.

Oliver, R. L. (1997). Satisfaction: a behavioral perspective on the consumer. New York: McGraw-Hill.Ordonez, L. D., Connolly, T., & Coughlan, R. (2000). Multiple reference points in satisfaction and fairness assessment.Journal of Behavioral Decision Making, 13, 329–344.

Owie, I. (1985). Incongruence between expected and obtained grades and students’ rating of the instructor. Journal ofInstructional Psychology, 12, 196–199.

Page, T. J., & Spreng, R. A. (2002). Difference scores versus direct effects in service quality measurement. Journal ofService Research, 4(3), 184–192.

Spreng, R. A., & Page, T. J. (2001). The impact of confidence in expectations of consumer satisfaction. Psychology &Marketing, 18(11), 1187–1204.

Spreng, R. A., MacKenzie, S. B., & Olshavsky, R. W. (1996). A reexamination of the determinants of consumersatisfaction. Journal of Marketing, 60(3), 15–32.

Sweeney, P. D., & McFarlin, D. B. (1993). Workers’ evaluations of the ‘ends’ and the ‘means:’ an examination of fourmodels of distributive justice and procedural justice. Organizational Behavior and Human Decision Processes, 55, 23–40.

Szymanski, D. M., & Henard, D. H. (2001). Customer satisfaction: a meta-analysis of the empirical evidence. Journal ofthe Academy of Marketing Science, 29(1), 16–35.

Tata, J. (1999). Grade distributions, grading procedures, and students’ evaluations of instructors: a justice perspective.The Journal of Psychology, 133(3), 263–271.

Tse, D. K., & Wilton, P. C. (1988). Models of consumer satisfaction formation: an extension. Journal of MarketingResearch, 25, 204–212.

Tyler, T. R., & Caine, A. (1981). The influence of outcomes and procedures on satisfaction with formal leaders. Journalof Personality and Social Psychology, 41(4), 642–655.

van den Bos, K., Lind, E. A., Vermunt, R., & Wilke, H. A. M. (1997). How do I judge my outcome when I do not knowthe outcome of others? The psychology of the fair process effect. Journal of Personality and Social Psychology, 72(5),1034–1046.

van den Bos, K., Wilke, H. A. M., Lind, E. A., & Vermunt, R. (1998). Evaluating outcomes by means of the fair processeffect: evidence for different processes in fairness and satisfaction judgments. Journal of Personality and SocialPsychology, 74(6), 1493–1503.

Vollmer, F. (1984). Expectancy and academic achievement. Motivation & Emotion, 8, 67–76.Wanous, J. P., Poland, T. D., Premack, S. L., & Davis, K. S. (1992). The effects of met expectations on newcomer

attitudes and behaviors: a review and meta-analysis. Journal of Applied Psychology, 77, 288–297.Westbrook, R. A., & Reilly, M. D. (1983). Value–percept disparity: an alternative to the disconfirmation of expectations

theory of consumer satisfaction. In R. P. Bagozzi, & A. M. Tybout (Eds.), Advances in consumer research (Vol. 10, pp.256–261). Ann Arbor: Association for Consumer Research.

Wicker, A. W., & Bushweiler, G. (1970). Perceived fairness and pleasantness of social exchange situations: two factorialstudies of inequity. Journal of Personality and Social Psychology, 15, 63–75.

Yi, Y. (1990). A critical review of consumer satisfaction. In V. A. Zeithaml (Ed.), Review of marketing (pp. 68–123).Chicago: American Marketing Association.

Authors’ biographies:Bennett Cherry is an Assistant Professor of Management and Organizational Behavior at California State UniversitySan Marcos. Ben’s research expertise is in service-delivery strategies, interpersonal-trust development in traditional andnontraditional environments, and humility in leadership.

394 Journal of Behavioral Decision Making

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)

Page 21: Grade expectations: the effects of expectations on fairness and satisfaction perceptions

Lisa D. Ordonez is an Associate Professor in the Management and Policy Department at The University of Arizona. Herresearch interests include judgment and decision making with applications in management and consumer behavior,regret, judgments of fairness, reference points, and the effects of sleep deprivation on medical decision making.

StephenW. Gilliland is a Professor of Management and Policy and the Vice Dean of the Eller College of Management atthe University of Arizona. Stephen’s research expertise is in the justice of management practices and policies. Throughthis merging of social, legal, and managerial issues, he has authored numerous articles and chapters.

Authors’ addresses:Bennett Cherry, California State University San Marcos, College of Business Administration, San Marcos, CA 92096–0001, USA.

Lisa D. Ordonez, Management and Policy Department, McClellaud Hall, University of Arizona, Tucson, AZ 85721-0108, USA.

Stephen W. Gilliland, Eller College of Management, McClellaud Hall, University of Arizona, Tucson, AZ 85721-0108,USA.

B. Cherry et al. Grade Expectations 395

Copyright # 2003 John Wiley & Sons, Ltd. Journal of Behavioral Decision Making, 16: 375–395 (2003)